5. CASE STUDY METHODOLOGY
5.3.1 Questionnaire instrument
As a ubiquitous method to assess personality, beliefs and attitudes (Wellington and Szczerbiński, 2007), the implementation of questionnaires is the most crucial stage of this case study analysis. The content of the questionnaire includes the three layers of social sustainability: basic needs, social networks and community development, which however, had been explicitly explained through plain questions about their feelings and attitudes towards their local neighbourhoods. The first important issue was the way of operating
questionnaires. Undoubtedly, different approaches have their own advantages and
disadvantages. The internet‐based questionnaire, although more efficient in terms of time and cost, may cause biases, as there may be a considerable proportion of population who has no access to the internet or no ability to answer in this way, especially amongst the elderly. The use of postal questionnaire can also save time, as it allows a parallel operation of the questionnaire among multiple cases. Some respondents may have potential weak reading abilities and comprehension abilities (Olsen, 2011), the questionnaire needs to be written with full and clear instructions. However, the problem of low response rate still commonly exists in postal questionnaires (Gilbert, 2008). The face‐to‐face questionnaire is reliable due to the researcher’s participation (Bryman, 2012), in which invalid answers can be to a large extent negated. However, it will inevitably increase the survey period and costs.
In the case study phase of this research, a factor that could fundamentally determine the questionnaire method was whether an on‐site survey could be allowed to be undertaken inside neighbourhoods. Due to the restriction by the gated residential management, some neighbourhoods did not approve researcher’s direct face‐to‐face survey request. As a result, a combination of multiple approaches had to be taken during the survey. Positive response rates were received when the face to face questionnaire survey was carried out, thanks to the assistance of volunteers and some neighbourhood committees. The use of a postal questionnaire was the remaining survey choice for the neighbourhoods where there were no face‐to‐face investigating opportunities.
104
Secondly, the large populations in local neighbourhoods required a sampling method to administer the questionnaire survey. The issue of sample size needs to be clarified from the outset. Sample size refers to the number of units chosen in the survey from which data was gathered (Lavrakas, 2008). However, so far there is no exact or straightforward answer for the appropriate size required for a questionnaire survey. Gathering real statistics from the population of a case study is the ground for deciding the sample size. Noticeably, a much larger sample size may not mean a higher precision in some questionnaire surveys (Shapiro, 2008). It could even significantly increase the possibility of sample error in the situation of a random sample. Nevertheless, some studies suggest an approximate 3%‐5% of the whole investigated population as an effective control of sample size (Frankfort‐Nachmias and Nachmias, 2008). Another way is to generate sample size according to estimate the standard error, confidence levels and interval. For this study, the actual population and spatial variations could make the sample size differ. However in practice, the actual response rates are often beyond the reach of many researchers (Wilson, 2013).
Thirdly, choosing correct sampling methods was also essential (Bernard, 2000). A feasible sampling method can be a sufficient estimation of a population and a reliable representation of its features (Bryman, 2012). It also needs to comply with the research’s focus and actual requirements. For better representing the real situations of local neighbourhoods, a spatially stratified random sampling method was adopted in this survey. This sampling process also had a hierarchical arrangement, approaching from blocks to households and then individuals in three stages. A Chinese urban neighbourhood is usually constituted by buildings, the basic geographical units. Some small neighbourhoods only have one or two buildings; their geographical variances may not be a significant issue for questionnaire sampling. Nonetheless, inner sub‐boundaries in a large neighbourhood usually exist that divide it into several sub‐areas or clusters. Residents living in different buildings may be far away from each other, and may have varied opinions due to larger spatial variances. Hence, the spatially stratified sample was the sampling solution. For individuals, decisions were made by residents inside of each household on who was
responsible for answering the questionnaire. However, a basic requirement was that only adult respondents were eligible to answer. It was always announced by the researcher and volunteers when conducting the survey.
Here is an example of the stratified random sampling approach used in this survey. A large‐sized neighbourhood with 3,000 households is split into three clusters. The sampling method may require 30 to 45 responses, which was estimated by proper survey confidential level and confidential interval. They were thus equally gathered from 10 to 15 samples in each group. Stratified samples were further applied in each cluster of buildings. Three samples were required for each of the five buildings in this cluster. For getting these samples, households were randomly selected by knocking doors, and one valid response was given representing this household. Occasionally, respondents were randomly chosen from the front of building entrances when the access to household inside buildings was not available. In extreme circumtance, a postal questionnaire was applied instead for few neighbourhoods where the researcher’s direct contact with participants was not authorised. The postal dispatching was also designed in a stratified order, with the respondents apportioned according to pre‐identified neighbourhood geographic units.
Lastly, the questionnaire content (Appendix 1) was specially developed, which had three parts and contained 26 questions. To make sure respondents understand why and what they need to answer with these questions, a clear illustration was made at the top of the questionnaire. Detailed explanations for certain questions were also given when necessary. Section A was the registration of some basic information of participants, in which useful social‐demographical data could also be collected. In Section B, the social sustainability assessment was delivered by essential questions explicitly discussing the elements under the framework ‘individual needs’, ‘social networks’ and ‘community development’. Section C was a policy indication test about the significance of social sustainability for residents and the value preference of different elements. Additional opinions on neighbourhood development, planning and governance were also welcomed but totally optional. Within
106
the content, numerous questions were designed using the ‘Likert scale’ five‐point measurement, which has been widely used for testing attitudes in questionnaire surveys (Croasmun and Ostrom, 2011). A middle‐value choice for neither agree nor disagree is also a common pattern in the ‘Likert scale’ (Addington‐Hall, 2007).
5.3.2 Observationinstrument
The observation method was another substantial component of the social sustainability assessment. Evaluating space quality and design, recording behaviour at specific times and places have often been adopted as visual tools during observation (Atkinson and Delamont, 2010). In this case study, the observation focuses on outdoor physical activities. To explicitly demonstrate the abstract concept human ‘interaction’, the approach of assessing ’outdoor activity’ have been developed by urban researchers (Gehl and Gemzoe, 2004). The outdoor activity contains different levels of actions, which were initially categorised into three types (Gehl, 1971): necessary activities (NA), optional activities (OA) and social activities (SA). Firstly, the NA includes those basic everyday actions under mandatory needs such as walking to work and shopping. But it mostly happens individually, rarely involves social interactions. Secondly, typical OA includes sitting, resting, playing, doing sports alone, babysitting and childminding. The OA, with a full range of vitality, is important because its occurrence means people are happy to relax inside their neighbourhoods. By contrast, when people do not like their environment or feel uncomfortable, the number of OA would decrease. Thirdly, the type of SA encompasses seeing, hearing, active contacts greetings, talking, conversations, children playing, demonstrating a strong social feature of a neighbourhood. Additionally, there are also some interesting cultural activities of SA in China. People in neighbourhoods often play traditional games such as cards or Mahjong in public spaces. Social activities undoubtedly represent strong social interactions and reinforce social ties among residents. The optional activities and social activities are recommended by many researchers (Biddulph, 2012, Zhang and Lawson, 2009) because of their ability to indicate strong social network connections. In this study, only optional and social activities are defined as valid outdoor activities for social interaction assessment.
Although the OA may be less meaningful than the SA in terms of high‐level activities, it still underpins social interaction to a certain degree.
In practice, recording outdoor activities are often implemented in empirical studies (Donald, 2011, Matan, 2011). However, a real issue for observation is how to record these activities systematically. The observation opportunity and time are often problematic. Due to the limitation of times and costs, observation in this research is confined to ‘observation for a certain period’. To avoid the hot outdoor environment in the summer of Shenzhen during which the activities might be hampered, the research chose to operate observations in October 2013 when the weather was relatively cool and comfortable. The representative recording time was designed between 4 pm and 6 pm. Most outdoor activities happened during this period when people come back from work and relax inside neighbourhoods before sunset. Although a small volunteer group were recruited, it is unlikely to complete the observations of all neighbourhoods in the same day at the same time. Greater manpower and workloads were also required for the observation taken place in some large sites. As a result, ‘acceptable equivalent periods’ were applied in the series of on‐site observations. Their dates and times were selected under similar weathers and similar hours (Appendix Table 3.2), assuming that the basic outdoor conditions for neighbourhood activities were equal and the comparison could be meaningful. The types of activities were coded by series of symbols and recorded on each neighbourhood map by the researcher and volunteers (Appendix Table 3.3). Additional translations and calculations of this raw data, which came from these sub‐categories inside the groups of OA and SA, were also necessary.
5.3.3 Interviewsanddocuments
Interviews were additionally undertaken with different stakeholders to understand their current roles and duties in the local level of neighbourhood governance structure. The community offices, neighbourhood developers, property management teams, and resident committees were the four major groups of interviewees. Much useful information was
108
gathered during these interviews with stakeholders, such as neighbourhood population data, renting household and immigrant data and local crime reports. Another purpose is to investigate whether there is an effective channel to achieve social sustainability in at the lowest neighbourhood scale. Semi‐structured interviews were applied to each type of groups, which followed certain topics and questions that were set as clear goals to achieve. However, interviewees were invited to introduce any related issues on their own, depending on their time and willingness to engage in discussions. From their responses and assertions, the perspectives of key stakeholders can be acknowledged and summarised. A series of coding of the questions and interviewees was continued as a sequential classification of their different positions and role‐plays (Appendix Table 3.1).
5.4
Ethicsoftheresearchsurvey
A basic guideline of ethical considerations in research is that no individual suffers any adverse consequences from participating in the study (Floyd J. Fowler, 2009). Whether a question is concerned with potentially sensitive or threatening information into content needs to be carefully considered during a survey. Although discussions on some planning defects might be embarrassing for some interviewees, the research’s objective of seeking suggestions for shaping more socially sustainable neighbourhoods still attracted many stakeholders’ interests. In general, improving social sustainability also contributes to China’s urban development process. During the survey, many participants did not refuse to participate following the introduction of these ideas. The researcher ensured that respondents agreed to participate with legal considerations and each participant therefore gave their informed consent. A part of preliminary research reports to the local planning bureau was sent in advance for a discussion. The discussions with local communities, the bottom organisational unit of the city, were also sought for a legal research purpose. The entire research survey was conduced under the approval and support of the local planning bureau, the planning research centre and other departments of local authorities.